Potential climate change impacts on the water balance of regional ...

Hydrol. Earth Syst. Sci., 16, 4581–4601, 2012 www.hydrol-earth-syst-sci.net/16/4581/2012/ doi:10.5194/hess-16-4581-2012 © Author(s) 2012. CC Attribution 3.0 License.

Hydrology and Earth System Sciences

Potential climate change impacts on the water balance of regional unconfined aquifer systems in south-western Australia R. Ali1 , D. McFarlane1 , S. Varma2 , W. Dawes1 , I. Emelyanova1 , and G. Hodgson1 1 CSIRO 2 CSIRO

Floreat Laboratories, Private Bag 5, Wembley, 6913, Western Australia Earth Science and Resource Engineering, 26 Dick Parry Avenue, Kensington WA 6151, Australia

Correspondence to: R. Ali ([email protected]) Received: 26 April 2012 – Published in Hydrol. Earth Syst. Sci. Discuss.: 24 May 2012 Revised: 16 November 2012 – Accepted: 16 November 2012 – Published: 4 December 2012

Abstract. This study assesses climate change impacts on water balance components of the regional unconfined aquifer systems in south-western Australia, an area that has experienced a marked decline in rainfall since the mid 1970s and is expected to experience further decline due to global warming. Compared with the historical period of 1975 to 2007, reductions in the mean annual rainfall of between 15 and 18 percent are expected under a dry variant of the 2030 climate which will reduce recharge rates by between 33 and 49 percent relative to that under the historical period climate. Relative to the historical climate, reductions of up to 50 percent in groundwater discharge to the ocean and drainage systems are also expected. Sea-water intrusion is likely in the PeelHarvey Area under the dry future climate and net leakage to confined systems is projected to decrease by up to 35 percent which will cause reduction in pressures in confined systems under current abstraction. The percentage of net annual recharge consumed by groundwater storage, and ocean and drainage discharges is expected to decrease and percentage of net annual recharge consumed by pumping and net leakage to confined systems to increase under median and dry future climates. Climate change is likely to significantly impact various water balance components of the regional unconfined aquifer systems of south-western Australia. We assess the quantitative climate change impact on the different components (the amounts) using the most widely used GCMs in combination with dynamically linked recharge and physically distributed groundwater models.

1

Introduction

Climate change affects temperature, rainfall and relative humidity and has a flow-on effect throughout the hydrological cycle (Lo´aiciga et al., 1996). There is evidence of a temperature increase of 0.74 ◦ C in the average surface temperature of the earth during the 20th century (IPPC, 2007; UNEP, 2007) and a further increase of 1.4 to 5.8 ◦ C by 2100 is projected due to greenhouse gas emissions (McCarthy et al., 2001). The climate change has impacted rainfall in many parts of the world (Ducci and Tranfaglia, 2008) and has caused significant reductions in a number of regions impacting on the availability of both surface and groundwater resources. There is evidence of decline in summer and autumn precipitation in drier northeastern regions of China since 1960 (Piao et al., 2010) and decline in mean annual rainfall in the southwest Western Australia and parts of the southern and eastern Australia during the second half of the 20th century (PMSEIC Independent Working Group, 2007). Climate change is likely to further impact rainfall in the future in many regions of the world and in some parts precipitation is projected to increase (w.g. van Roosmalen, 2007, 2009). Increasing or decreasing precipitation and the resulting increase or decrease in groundwater recharge or discharge may also deteriorate dependent or associated ecosystems and agricultural production (Aldous et al., 2011; Barron et al., 2012; Hinsby et al., 2012; Jeppesen et al., 2009; Olesen et al., 2007; Sonnenborg et al., 2012). Global climate models (GCM) are used for projecting future climates. Recent advances in modelling and improved understanding of the physical processes of climate systems have made projections of future climate more reliable

Published by Copernicus Publications on behalf of the European Geosciences Union.

4582

R. Ali et al.: Potential climate change impacts on the water balance of unconfined aquifers

(Christensen et al., 2007; Pirani, 2011; Taylor et al., 2012) and simulation results from downscaled global circulation models have been used for regional studies of the impact on the hydrological cycle by integrated hydrological models (e.g. van Roosmalen et al., 2007, 2009; Goderniaux et al., 2009, 2011) including water quality issues (Sonnenborg et al., 2012). Both the temperature rise and rainfall decline due to climate change affects various components of the groundwater balance by directly and indirectly affecting multiple factors (Zagonari, 2010) such as groundwater recharge (Ng et al., 2010) due to rainfall decline, potential evapotranspiration due to temperature and vapour pressure deficits rise, and groundwater discharge to drains due to groundwater levels decline. It can increase seawater intrusion risks in coastal aquifers and affect inter-aquifer leakage rates and flow direction. Although the research on climate change effects on groundwater systems is relatively scarce (Marshall and Randhir, 2007) it is rapidly increasing (Barthel et al., 2011; van Roosmalen et al., 2007, 2009; Goderniaux et al., 2011; Sonnenborg et al., 2012). Some studies have either fully or partly addressed the effects of recent past and projected climate change on groundwater resources using various hydrological, regression, modelling and isotope techniques (Scibek and Allen, 2006; Isaar, 2008; Ducci and Tranfaglia, 2008; Polemio and Casarano, 2009; Sinha and Navada, 2008). Only few studies have applied truly integrated hydrological models that are physically-based and spatially distributed both on the surface and in the subsurface (Goderniaux et al., 2009). Often variable results are produced (Jiang et al., 2007) due to simplistic assumptions made in climate projections and representation of physical processes of the hydrological system. A reliable estimate of recharge is the first requirement in estimating the impact of climate change on groundwater systems because it represents the connection between atmospheric, surface and sub-surface processes and is impacted by many factors including changed precipitation and temperature regimes, coastal flooding, urbanisation and surface sealing, woodland creation and rotation changes (Holman, 2006). Various simple water balance techniques and models are available to estimate recharge (Aguilera and Murillo, 2009; Barr et al., 2003; Herrera-Pontoja and Hiscock, 2008; Sanford, 2002). It can be used as input to numerical groundwater models which distribute this recharge into various components of the water balance, or a recharge model can be dynamically linked with the groundwater models which estimate recharge at each stress period depending on variations in climate and land surface conditions. Input from recharge models can be used in groundwater models to study the impacts of climate change on groundwater systems. Scibek and Allen (2006) and Woldeamlak et al. (2007) used physically-based distributed recharge models to estimate annual recharge and used this in MODFLOW model to simulate groundwater in Grand Forks in south central British Columbia and in Grote-N Belgium, respectively. Hydrol. Earth Syst. Sci., 16, 4581–4601, 2012

Various recharge and groundwater models have been used in Australia to study the impacts of climate change on future groundwater resource availability in the Murray-Darling Basin, Northern Australia, Tasmania and south-west Western Australia (Ali et al., 2010, 2012; Crosbie et al., 2010; CSIRO, 2008, 2009; Post et al., 2011). Most studies used recharge as a direct input in groundwater models. Our study dynamically linked a recharge model that takes account of variations in climate, land surface condition and groundwater depth with MODFLOW based groundwater models. By dynamically linking a recharge model with a groundwater model the effect of depth to water table on recharge was also taken into account. Quantification of storage change and discharge changes by simultaneously considering unsaturated and saturated processed based models is an important step forward in this field of study. The main objectives of this study were to dynamically link a recharge model with MODFLOW based groundwater models and simulate the potential climate change impacts on the water balance of the coastal plain unconfined aquifers of south-western Australia.

2

Description of study area

The study area is located between Gingin to the north and Augusta to the south of Perth (Fig. 1). For this study the area was divided into three parts: the Central Perth Basin, the Peel-Harvey Area and the Southern Perth Basin, all of which are part of the Perth Basin (Fig. 1). The study area covers about 20 000 km2 and is located in one of the highest rainfall parts of south-western Australia. It includes all fresh, brackish and marginal groundwater resources near the coast. Inland groundwater supplies are either limited or too saline for most domestic, irrigation and industrial uses and were therefore excluded from this assessment. The study area has over 80 percent of Western Australia’s population and accounts for over half of the horticultural production of the state (van Gool and Runge, 1999). Groundwater is a major source of water in the study area. The Perth Basin comprises the flat sandy Swan and Scott coastal plains, and more elevated and clayey plateaux such as the Blackwood in the Southern Perth Basin and Dandaragan in the Central Perth Basin (Fig. 1). 2.1

Climate

The study area has a Mediterranean type climate, with mean annual rainfall of between 500 and 1200 mm, up to 80 percent of which occurs between May and October. Temperatures are also at their lowest during this period, making the rainfall more effective in terms of producing runoff and recharge. There is a strong south-west to north-east gradient in rainfall (Fig. 2) with the highest rainfall in south-west coastal parts and along the Darling Range east of the Darling www.hydrol-earth-syst-sci.net/16/4581/2012/


4583

111 Fig. 1. Map of the study area and main 112 Fig.tectonic 1 Mapsubdivisions. of the study area and main tectonic subdivisions

113

2.1 Climate

peri-urban parts of Perth, and in south-western coastal areas Scarp. The mean annual areal potential evapotranspiration at Myalup, Jindong and Margaret River. (APET; Morton, 1983) varies from 1555 mm in the north to 114mm The study area has mean a Mediterranean typedeficit climate, 1260 in the south. When annual rainfall is with mean annual rainfall of between 500 and 1200 calculated by subtracting APET from rainfall, almost all of 2.3October. Hydrogeology 115 mm, up to 80 percent of which occurs between May and Temperatures are also at their the study area has a negative moisture balance.

116

lowest during this period, making the rainfall more effective in termsPerth of producing runoff 150 and km recharge. The Central Basin extends north and 70 km

of Perth and comprises the Swan Coastal Plain and 2.2 coveris a strong south-west to north-east gradient in south 117Land There rainfall (Fig. 2) with thePlateau. highestThe rainfall in main aquifers are southern Dandaragan three the Superficial, Leederville and Yarragadee. Locally impor118 south-west coastal parts and along the Darling Rangetant east of the Darling Scarp. The mean annual aquifers include the Parmelia, Rockingham, Mirrabooka Major land covers in the study area include dryland agriculand Poison Hill 1555 aquifers 4).north The lithology and hydraulic ture, native vegetation, pine plantation and urban (Fig. 3). 119 areal potential evapotranspiration (APET; Morton, 1983) varies from mm(Fig. in the to 1260 mm conductivity of geological units are listed in Table 1. The Over 88 percent of the study area is covered by either native vegetation or dryland (43 percent). sub-surface geology and sub-cropping for the Cen120 in(45 thepercent) south. When meanagriculture annual rainfall deficit is calculated by subtracting APET from rainfall,strata almost Pine plantations cover 2 percent of the area. Horticulture, tral Perth Basin (Fig. 4a) show the degree of connectivity 121 at $166 all of million the study area has a negative balance. valued annually covers less thanmoisture 1 percent of with underlying aquifers such as the Leederville Aquifer. the study area but makes up 33 percent of total agricultural The unconfined Superficial Aquifer is mainly clayey (Guildproduction (van Gool and Runge, 1999). Intensive urban and ford Formation) in the east near Gingin Scarp and contains commercial buildings cover about 6 percent of the study area various sandy formations such as Safety Bay Sand, Becker and occur in the Perth region and, to a lesser extent, around Sand and Tamala Limestone towards the Indian Ocean in the Bunbury and Busselton. west. Its average thickness is about 30 m. The Pinjar, WanThe main areas under native vegetation occur east of the neroo and Mariginiup members of the Leederville FormaDarling Fault in the Darling Ranges, along the South Coast tion directly underlie the superficial formations in the northand north of Perth. Pine plantations occur in the Gnangara east and south and the Osborne Formation in the middle area north of Perth, near Myalup east of Lake Clifton and on (Fig. 4b). The Yarragadee is too deep to underlie the Superficial Aquifer in most areas but it does subcrop in the extreme the Blackwood Plateau. Cleared areas used for dryland cropping and grazing mainly occur on the Swan Coastal Plain. Irnorth and underlies the Wanneroo Member between Guilderrigated areas occur in the Harvey and Preston areas (Fig. 3). ton and Gingin, an important intake area of this aquifer. The Self-supplied horticultural areas, i.e. using water extracted Leederville Aquifer is often separated from the Superficial from farmer-owned wells or dams, occur around Gingin, in Aquifer by the Kardinya Shale (Osborne Formation) and the www.hydrol-earth-syst-sci.net/16/4581/2012/

5

Hydrol. Earth Syst. Sci., 16, 4581–4601, 2012

4584


122 Fig. 2. SpatialFig. distribution of distribution mean annualofhistorical (1975 to 2007) rainfall, evapotranspiration and rainfall deficit (rainfall 123 2 Spatial mean annual historical (1975 areal to 2007) rainfall, areal(APET) evapotranspiration less areal potential evapotranspiration) across the study area. 124 (APET) and rainfall deficit (rainfall less areal potential evapotranspiration) across the study area

2.2 Land cover

upward hydraulic gradients and confining beds are absent. Recharge rates vary depending on rainfall, lithology, depth to water tablenative and topographic and landand cover. Ground126 Major land covers in the study area include dryland agriculture, vegetation,gradient pine plantation water discharge is to the ocean, to natural and engineered drainages and tonative coastal lakes. Large losses occur through 127 urban (Fig. 3). Over 88 percent of the study area is covered by either vegetation (45 percent) or evaporation from wetlands and in areas with shallow water 128 dryland agriculture (43 percent). Pine plantations cover 2 percent of the area. Horticulture valued at towards the tables. Groundwater flow is usually ,westwards coast. The Gnangara Mound north of Perth and the Jandakot 129 $166 million annually covers less than 1 percent of theMound study area of total southbutofmakes Perth up are33 thepercent main flow systems. Further details about hydrogeology and confined aquifer systems of 130 agricultural production (Gool and Runge, 1999). Intensive urban and commercial buildings cover about the Central Perth Basin can be found in CSIRO (2009) and Davidson 131 6 percent of the study area and occur in the Perth region and, to(1995). a lesser extent, around Bunbury and The Peel-Harvey Area extends between the coastal cities 132 Busselton. of Mandurah and Bunbury and the Darling Scarp (Fig. 5a). The major aquifers are the Superficial, Leederville and Cock133 The main areas under native vegetation occur east of leshell the Darling theand Darling GullyFault (Fig.in5b) their Ranges, lithologyalong and hydraulic conductivity is listed in Table 1. The Superficial Aquifer extends 134 the South Coast and north of Perth. Pine plantations occur in the Gnangara area north of Perth, near over the entire region and consists of the Bassendean Sands and the Ascot, areas Guildford formations. Its thick135 Myalup east of Lake Clifton and on the Blackwood Plateau. Cleared usedand for Yoganup dryland cropping ness ranges from 20 to 30 m. The Superficial Aquifer is un136 and grazing mainly occur on the Swan Coastal Plain. Irrigated areas occur in Harveyaquifer and Preston derlain by a number of the confined systems including the Leederville, Cockleshell Gully and Yarragadee. 137 areas (Fig. 3). Self-supplied horticultural areas, i.e. usingThe water extracted farmer-owned or Recharge water table isfrom generally shallow inwells the area. Fig. 3. Major land cover types in the study area (using satellite immainly occurs through rainfall infiltration but some is re138 dams, in occur Gingin, in peri-urban of Perth, and in south-western coastal areas at agery from types 2005). 3 Major land cover the around study area (using satelliteparts imagery from jected2005) due to shallow water tables. Groundwater flow is towards the Indian Ocean in the west and discharges into the 139 Myalup, Jindong and Margaret River. natural and engineered drainages, wetlands and ocean. Furdrogeology ther details on the hydrogeology and confined aquifer sysLeederville is often separated from the Yarragadee Aquifer 140 tems of the Peel-Harvey Area are given in CSIRO (2009) by the South Perth Shale (Fig. 4b). erth Basin extends 150bykm northinfiltration and 70 km south of Perth and the Swan and URS (2009a). Recharge rainfall often decreases from thecomprises The Southern Perth Basin lies between the Darling and coast to the east because of decreasing rainfall and sediments and southern Dandaragan Plateau. three can main aquifers are the Dunsborough Superficial, Faults and includes the easterly Bunbury becoming clayey in the east.The Recharge also occur from Trough and westerly Vasse Shelf which are separated by upward leakage from underlying formations where there are d Yarragadee. Locally important aquifers include the Parmelia, Rockingham, Mirrabooka

125

6

Hydrol. Earthlithology Syst. Sci., 16,hydraulic 4581–4601, 2012 www.hydrol-earth-syst-sci.net/16/4581/2012/ l aquifers (Fig. 4). The and conductivity of geological units are listed

e sub-surface geology and sub-cropping strata for the Central Perth Basin (Fig. 4a)

hydrogeology and confined aquifer systems of the Central Perth Basin can be found in CSIRO (2009) R. Ali et al.: (1995). Potential climate change impacts on the water balance of unconfined aquifers and Davidson

4585

4. Sub-surface geology of the Central Perth Basin (a) and geological cross-section A-A’ (b) (adapted from Davidson and Yu, 2008). Fig. 4Fig. Sub-surface geology of the Central Perth Basin (a) and geological cross-section A-A’ (b)

(adapted from Davidson Yu,Major 2008) the Busselton Fault (Fig. and 6a, b). aquifers include the Superficial, Leederville and Yarragadee (Baddock, 2005) (Fig. 6b); their lithology and hydraulic conductivities are listed in Table 1. The Lesueur Sandstone, Cockelshell Gully and Sue Coal Measures are minor aquifers. The Superficial Aquifer extends over the Swan Coastal and Scott Coastal plains. The Superficial Aquifer material is more uniform and sandy under the Scott Coastal Plain than under the Swan Coastal Plain, however the aquifer can be locally confined by a ferruginous cemented layer in the lower parts of the formation (Strategen, 2005). Its thickness is often less than www.hydrol-earth-syst-sci.net/16/4581/2012/

8

10 m but is up to 200 m under the coastal sands. Three confining beds occur within the basin: the middle clay unit of the Parmelia Formation, the Bunbury Basalt and the upper Mowen/Quindalup Member of the Leederville Formation (Fig. 6b and Table 1). The middle clay unit is about 35 m thick and extends over the Bunbury Trough beneath the Blackwood Plateau (Strategen, 2004). The overlying and underlying formations are the Bunbury Basalt or the Leederville Aquifer and the Yarragadee Aquifer, respectively. The Lesueur Sandstone extends throughout the Southern Perth Basin. The Cockelshell Gully Formation is overlain Hydrol. Earth Syst. Sci., 16, 4581–4601, 2012

the hydrogeology and confined aquifer systems of the Peel-Harvey Area are given in CSIRO

09), URS (2009a) and ANRA (2009). 4586


Fig. 5. Sub-surface geology of the Peel-Harvey Area (a) and geological cross-section B-B’ (b) (adapted from URS, 2009).

.5 Sub-surface geology of the Peel-Harvey Area (a) and geological cross-section B-B’ (b) (adapte

by the Yarragadee Formation which is separated into the fined aquifer systems of the Southern Perth Basin are given lower Yarragadee 4, Yarragadee 3, Yarragadee 2 and upper in CSIRO (2009) and Strategen (2005). m URS, 2009) Yarragadee 1 (Fig. 6b). 2.4 Historical groundwater use and future demand On both Swan Coastal and Scott Coastal plains the Supere Southern lies between Darling ficial Perth Aquifer isBasin mainly recharged by rainfall the infiltration with and Dunsborough Faults and includes the easter Western Australia is faced with a scientifically complex chalsome upward leakage from the underlying Leederville or lenge as the state relies heavily on groundwater systems Yarragadee aquifers near Bunbury within the Swan Coastal nbury Trough and westerly Vasse Shelf which are separated byto the Busselton Fault (Fig. which are difficult quantify, due to their complex geology6a, 6b). Plain and in the centre of the Scott Coastal Plain. The groundand hydrogeology, so that resource availability often requires water discharge is to the natural drainages, wetlands, ocean sophisticated measurements (DoW, 2008a). Groundwater use or to include the underlying and Yarragadee aquifers.and Yarragadee (Baddock, 2005) (Fig. 6b); jor aquifers theLeederville Superficial, Leederville their has been sharply increasing over 30-yr across all sectors in The main groundwater flow direction is to the north toWestern Australia. The annual abstraction from aquifers in wards the Indian Ocean with some southerly flow towards the the Central Perth Basin (Figs. 1 and 7a) trebled from 200 GL Southern Ocean. The Leederville and the Yarragadee outcrop in 1985 to nearly 600 GL in 2007. Abstraction from the Suover extensive areas and receive direct recharge from rainfall perficial Aquifer in the Peel-Harvey Area (Figs. 1 and 7b) infiltration. Further details about hydrogeology of the conhas increased from about 0.3 GL in 1994 to about 20 GL by

9

Hydrol. Earth Syst. Sci., 16, 4581–4601, 2012

www.hydrol-earth-syst-sci.net/16/4581/2012/


4587

Table 1. Lithology and estimated horizontal hydraulic conductivity of geological units in the study area (Davidson and Yu, 2006; Strategen, 2005; Sun, 2005; URS, 2009). Formation

Lithology

Horizontal hydraulic conductivity (m day−1 )

Central Perth Basin (PRAMS) and Peel-Harvey Area (PHRAMS) Safety Bay sand

Sand and shelly sand

10 to 15

Becker sands

Fine to medium grained quartz and skeletal sand with lenses of calcareous clay

8

Tamala limestone

Limestone with weathered beds of sand, marl and mud present at the base

100 to1000 for limestone 7 to 35 for sandy beds

Bassendean sand

Sand and subordinate silt and clay

10 to 50

Guildford Fm

Clay with subordinate sand and gravel